INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

ENDOTHELIAL CELLS exposed to blood flow are thought to have mechanosensors that convert shear stress-related mechanical forces in the plasma membrane into specific cellular signaling system(s) (4, 7, 10, 14). This paradigm has been studied predominantly with endothelial cells subjected to increasing levels of shear, whereas the effect(s) of decreased shear is relatively poorly characterized or understood. In ischemia (e.g., a vessel occlusion), flow cessation is generally followed by hypoxia and cultured cells subjected to hypoxia have been used as a model for ischemia (28, 29). However, the response to hypoxia alone cannot adequately model the endothelial response to the mechanical component of ischemia. Furthermore, endothelial cells are normally preconditioned to flow and their ability to sense and respond to ischemia may differ from the responses of static cells that are known to lose certain endothelial-specific features (e.g., caveoli and cytoskeleton organization) when removed from the circulation (10, 11, 21).

We previously used imaging techniques in an isolated rat lung model, continuously ventilated to maintain oxygenation, to separate ischemia from hypoxia and to detect pulmonary endothelial responses to loss of shear stress in situ. Based on measurements with fluorescent dyes, flow cessation resulted in depolarization of the endothelial plasma membrane, generation of reactive O₂ species (ROS) and increase in intracellular Ca²⁺ concentration ([Ca²⁺]_i) (1, 3, 25). To isolate the response by endothelial cells from that due to other cell types in the tissue and to assess this paradigm in a more generic endothelial cell type, we expanded these studies to an in vitro system of endothelial cells preadapted to flow in artificial capillaries. This latter study showed that flow cessation (simulated ischemia) results in increased ROS generation, activation of nuclear factor- (NF-) and activator protein-1, and subsequent cellular proliferation (27). However, this in vitro artificial capillary system could not be used to detect and quantitate the initial responses to ischemia in real time kinetics because the artificial capillaries are opaque and direct observation of cells by microscopy or spectroscopy is not possible. Furthermore, trypsinization was required to detach cells before their study, necessitating a 5- to 10-min delay between the ischemic stimulus and subsequent cellular examination (27).

The goal of the present study was to characterize the initial molecular events after flow cessation in flow-adapted endothelial cells. We designed a laminar-flow, parallel-plate chamber that can be used for flow adaptation and also for the continuous spectroscopic study of an endothelial cell monolayer. This system allowed us to study the initial cellular metabolic events after onset of simulated ischemia. Our results indicate that flow-adapted endothelial cells, but not those cultured under static conditions, respond to flow cessation by an immediate membrane depolarization and powerful oxidative burst, followed by increase in [Ca²⁺]_i and activation of Ca²⁺/calmodulin-dependent nitric oxide (NO) synthase.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Materials. Ferricytochrome c (cyt c; from horse heart), HEPES, Cu-Zn superoxide dismutase (EC1.15.1.1) (SOD) from bovine erythrocytes, catalase (EC1.11.1.6) from bovine liver, and EGTA were purchased from Sigma (St. Louis, MO). 4,5-Diaminofluorescein (DAF-2) diacetate was purchased from Calbiochem (San Diego, CA). The phosphorescence probe, PdP1, was a gift from Dr. S. Vinogradov, Biochemistry and Biophysics Department of the University of Pennsylvania. Diphenyleneiodonium chloride (DPI) was purchased from ICN Biomedicals (Costa Mesa, CA). Bis-(1,3-dibutylbarbituric acid)trimethine oxonol (bis-oxonol), 2',7'-dichlorodihydrofluorescein (H₂DCF) diacetate, fura 2-acetoxymethyl ester (AM), N-6-aminohexyl-5-chloro-1-naphthalene sulfonamide (W-7), and N^G-nitro-L-arginine methyl ester (L-NAME) were purchased from Molecular Probes (Eugene, OR). Eagle's minimal essential medium (MEM) and other cell culture reagents were purchased from GIBCO BRL (Gaithersburg, MD). Fetal bovine serum (FBS) was purchased from Hyclone (Logan, UT). Pronectin F was purchased from Protein Polymer Technologies (San Diego, CA). Bovine pulmonary artery endothelial cells (BPAEC) were purchased from the American Type Culture Collection (Manassas, VA) at passage 15; we have shown previously that these cells show a similar response to ischemia as BPAEC at passage 3-4 (27).

Laminar flow chamber. A flow chamber with parameters similar to those of a parallel-plate chamber (19) was designed for spectroscopic study of a confluent monolayer of BPAEC after adaptation to laminar flow. BPAEC adherent to a 12.5 × 25 mm quartz or plastic slide (Aclar, Allied Signal; Morristown, NJ) were placed inside the chamber and perfused by using a peristaltic pump (Harvard Apparatus; South Natick, MA) (Fig. 1A). A rubber ring gasket was used to channel the flow. Observation of dye added to the perfusate indicated that laminar flow conditions were achieved with flow rates between 0.5 and 9 ml/min. The volume of the chamber perfusion area was 28 µl, and the total volume of the perfusion circuit was ~20 ml. The pump, perfusate reservoir, and chamber were connected with Tygon tubing (1/32 in. ID, 5/32 in. OD, Cole-Parmer Instrument) (Fig. 1). Temperature control (at 37°C) was maintained with a water circulator (model 1104, VWR Scientific). The 10 × 10 mm chamber was made of stainless steel (Fig. 1, B-D) with an insulated permanent quartz window and dimensions to fit the standard spectrophotometer/spectrofluorometer sample holder. Its geometry (optical path ~0.35 mm, 45° angle orientation of plastic slide with adherent cells to the direction of excitation beam) was designed for use in a UV-VIS spectrometer or with a light microscope. The plastic slide transmitted ~95% of light at all wavelengths between 220 and 800 nm.

View larger version (36K): [in this window] [in a new window]   Fig. 1.   Details of the perfusion flow chamber (FC) used to adapt cells to flow and to study the initial events after cessation of flow. A: scheme of the perfusion circuit showing a slide with confluent monolayer of bovine pulmonary artery endothelial cells (BPAEC) in the laminar FC. The FC and perfusate reservoirs were placed in a temperature-controlled bath. B: digitized photograph showing assembled (top) and disassembled (bottom) FC. C: diagonal cross section of flow area of the FC. IN and OUT indicate the perfusion flow path. D: top view of horizontal cross section of flow area.

Cell culture and simulated ischemia. Quartz or plastic slides were sterilized with 70% ethanol, air dried in a tissue culture hood, treated for 20 min at 37°C with 1 ml of a 50 µg/ml solution of Pronectin F, and washed twice with MEM. BPAEC were grown on slides placed in a 35-mm tissue culture dish (two identical slides per dish). BPAEC in MEM supplemented with 15% FBS (Hyclone) and 100 U/ml penicillin, 100 µg/ml streptomycin, 1 mM sodium pyruvate, 100 µM of nonessential amino acid solution, and 10 mM HEPES, pH 7.4 were plated at 1 × 10⁶ cells per dish (2 slides) and allowed to grow for 48 h until they became fully confluent. The number of cells per slide at confluence in the perfusion area was ~1 × 10⁵. For adaptation to flow, a slide with adherent cells was perfused in the flow chamber with growth medium supplemented with 25 mM HEPES (pH 7.4) at 37°C generally for 24 h at an estimated shear stress at the cell surface of 5 dyn/cm². A similarly prepared slide was maintained under static culture conditions for the same duration (called static cells).

Spectroscopic measurements. Superoxide anion radical (O·) generation was evaluated by measurement of absorbance of cyt c (13) using a Beckman DU 640B (Beckman Instruments; Fullerton, CA) dual beam spectrophotometer (3 nm slits). For detection of O·, cyt c was added to perfusate to 150 µM final concentration. Reduction of cyt c by O· was measured by change in absorbance at 550 nm. Spectra were recorded at 30-s intervals during simulated ischemia or in the real-time kinetic mode with resolution of one point each 15 s.

Fluorescence microscopy. Slides with flow-adapted BPAEC before or after simulated ischemia were imaged with an epifluorescence microscope with ×100 objective (Nikon Diaphot TMD) and equipped with an optical filter changer (Lambda 10-2, Sutter Instrument). Excitation was accomplished with a mercury lamp with narrow bandpass filter (FITC 485/10), triple-band dichroic mirror (D/F/R-BS&M, Chroma Technolog; Brattleboro, VT), and a narrow bandpass filter (535/40 transmission/half bandwidth, in nm) for emitted light. Images were acquired during a 500-ms exposure with a computer-controlled cooled CCD camera (MicroMAX, Princeton Instruments; Princeton, NJ) using graphics control software (Metamorph Imaging System, Universal Imaging; West Chester, PA). After fluorescence images were acquired, matching phase-contrast images were taken from the same area.

Data analysis. Results are expressed as means ± SE. Curves were fit to original data using KaleidoGraph V.3.0.2 (Synergy Software; Reading, PA) for Macintosh computer using the smoothing procedure, linear or polynomial fit. Iterations were repeated until the correlation coefficient (R) for the fit exceeded 0.9.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Adaptation of a confluent monolayer of BPAEC to laminar flow as described above (shear stress at 5 dyn/cm² for 24 h) resulted in the alignment of cells to the direction of flow with reversion to their original shape during an additional 24-h incubation under static (no-flow) culture conditions (Fig. 2). Similar flow-mediated changes in endothelial cell alignment to laminar flow have been reported previously in other experimental systems in vitro (11, 22). This result indicates that our flow chamber was satisfactory for flow adaptation of endothelial cells.

View larger version (73K): [in this window] [in a new window]   Fig. 2.   Phase microscopy of confluent BPAEC in cell culture on a plastic slide. A: cells were cultured for 24 h and observed before exposure to laminar flow. B: same cells after exposure to laminar flow with shear stress of 5 dyn/cm2 for 24 h. Cells have aligned in the flow path. C: same cells after incubation for an additional 24 h under no flow (static) conditions.

Our previous studies using a membrane potential-sensitive fluorescent probe, bis-oxonol, in the oxygenated perfused rat lung preparation revealed that simulated ischemia caused endothelial plasma membrane depolarization (3). In the present study, flow-adapted BPAEC showed a rapid (within 15 s after onset of ischemia) increase in bis-oxonol fluorescence after cessation of flow compatible with plasma membrane depolarization, and a rapid decrease of bis-oxonol fluorescence after start of reperfusion compatible with repolarization (Fig. 3). There was no change in bis-oxonol fluorescence during simulated ischemia with BPAEC that had not been flow adapted (not shown).

View larger version (23K): [in this window] [in a new window]   Fig. 3.   Cell membrane potential as indicated by bis-oxonol fluorescence. Effect of simulated ischemia (stop flow) and reperfusion (start flow) on bis-oxonol fluorescence in a monolayer of flow-adapted (5 dyn/cm2, 24 h) BPAEC. Cells were prelabeled with 0.02 µM bis-oxonol and fluorescence was monitored at 520 nm (490 nm excitation). An increase in fluorescence indicates plasma membrane depolarization.

Cells in the chamber are oxygenated by perfusate flow. Therefore, it is expected that PO₂ in the medium and cells will decrease with ischemia as a result of O₂ utilization reflecting cellular metabolic activity. In this respect, our system is a model for ischemia in systemic blood vessels. Cellular PO₂ was monitored with the phosphorescent probe PdP1 that localizes extracellularly and displays increasing phosphorescence with decreasing perfusate PO₂. A detectable increase in the phosphorescence signal was noted at 4-5 min after flow cessation in the flow-adapted monolayer of BPAEC, indicating that PO₂ had decreased to about 45 mmHg (Fig. 4). Saturation of perfusate with 100% O₂ resulted in approximately fivefold prolongation (~20 min) in the time required for detection of increased phosphorescence compatible with the expected approximately fivefold increase in perfusate O₂ content. Cells cultured under static (no flow) conditions showed detectible phosphorescence after ~16 min of ischemia (Fig. 4), indicating a rate of O₂ consumption that was ~25% of that seen with flow-adapted cells.

View larger version (22K): [in this window] [in a new window]   Fig. 4.   Medium PO2 as indicated by phosphorescence intensity of the extracellular probe PdP1. Phosphorescence during simulated ischemia with BPAEC is expressed in arbitrary units. The initial increase in phosphorescence above baseline indicates PO2 ~45 ± 3 mmHg, whereas the upper plateau reflects essentially zero PO2 (anoxia). With flow-adapted cells perfused with air-saturated buffer, phosphorescence was undetectable for 4-5 min after flow cessation but then rose rapidly and reached a plateau at ~8 min. When flow-adapted cells were perfused with buffer saturated with O2, phosphorescence with ischemia remained undetectable for ~20 min and then rose rapidly to a plateau at ~25 min. With static (nonflow adapted) cells perfused with air-saturated buffer, phosphorescence remained at baseline for ~17 min after ischemia and then increased to reach saturation at ~37 min.

The high rate of O₂ consumption by the flow-adapted cells is compatible with a powerful oxidative burst triggered by ischemia as indicated by our previous studies in the perfused rat lung (29). Cyt c was used to detect extracellular O· generation by the cells. A spectral scan during ischemia demonstrated increasing absorbance maximal at 550 nm, indicating reduction of cyt c (Fig. 5A). Sodium dithionite was added to the perfusate at the end of each experiment to exclude possible saturation of the signal. Absorbance at 550 nm started to increase within 15 s after flow cessation in flow-adapted cells and reached saturation at ~4 min (Fig. 5B). More precise timing within the initial 15-s period was not possible. Similar reduction of cyt c ischemia was observed when the external cyt c concentration was 75, 100, or 150 µM (data not shown); for subsequent experiments, 150 µM was used to ensure an excess of the detector. No change in absorbance with simulated ischemia was observed with cells cultured under static conditions (Fig. 5B). Reduction of cyt c was diminished by the addition of SOD to the perfusate and was abolished by preincubation with 20 µM DPI, an inhibitor of flavin-dependent oxidases. Addition of catalase enhanced both the rate and amplitude of cyt c reduction during simulated ischemia, suggesting that O· produced with ischemia is partially dismutated to H₂O₂ that oxidizes cyt c and thus attenuates the increase in absorbance. A similar effect of O· dismutation on cyt c absorbance has been observed in experiments with polymorphonuclear leukocytes (24). There was no effect on cyt c reduction by preincubation with 1 mM L-NAME for 30 min (data not shown). To evaluate the effect of diminishing perfusate PO₂ on the rate of O· generation, the response to simulated ischemia was measured after saturation of perfusate with 100% O₂. Under these conditions, there was a linear increase of the cyt c absorbance at 550 nm that continued for at least 5 min after cessation of perfusate flow (Fig. 5B). Thus the decreasing rate of cyt c reduction beginning at about 1.5 min under control conditions likely reflects the increasing limitation of O₂ availability for O· generation during continuous ischemia.

View larger version (23K): [in this window] [in a new window]   Fig. 5.   Extracellular superoxide anion radical (O·) production by BPAEC during simulated ischemia. A: effect of simulated ischemia on absorbance spectra of 150 µM ferricytochrome c (cyt c) added to the perfusate of flow-adapted BPAEC (24 h at 5 dyn/cm2). Spectra were recorded every 30 s after the cessation of flow and showed increasing absorbency at 550 nm with successive scans. The top spectrum was generated by an excess of sodium dithionite added for complete reduction of cyt c. The increase of specific absorbance at 550 nm relative to the isosbestic point at 540 nm indicates reduction of cyt c by O·. B: same conditions as shown in A but showing absorbance at 550 nm as a function of time after stopping flow. Conditions are the following: cells that were not flow adapted (static); flow-adapted cells under normoxia (control, air); preperfusion of flow-adapted cells with perfusate saturated with O2 (control, O2); preperfusion with diphenyleneiodonium chloride (DPI) (20 µM); and addition of superoxide dismutase (SOD, 20 units) or catalase (25 units) to flow-adapted cells. Results are means ± SE for three independent experiments for each condition. C: effect of shear stress magnitude during 24-h adaptation to flow on subsequent O· generation by BPAEC during simulated ischemia as in B. The shear stress during adaptation is indicated as dyn/cm2. OD, optical density.

These studies of simulated ischemia were done with cells adapted for 24 h to a shear stress of 5 dyn/cm². To study the effect of shear stress during adaptation on subsequent O· generation during simulated ischemia, BPAEC were subjected to a level of shear stress between 1 and 10 dyn/cm² during the 24-h adaptation period. The rate of O· generation during simulated ischemia increased progressively with increasing values of shear stress during the prior adaptation period reaching apparent saturation at ~5 dyn/cm² (Fig. 5C).

The cyt c experiments indicated accumulation of O· in the extracellular compartment. To evaluate whether ROS accumulated intracellularly during simulated ischemia, the fluorophore H₂DCF trapped inside flow-adapted cells was used (2, 27). Within 30 s after onset of simulated ischemia there was a linear increase in fluorescence in flow-adapted cells, indicating H₂DCF oxidation to DCF compatible with increased intracellular ROS. The DCF signal reached a plateau in ~5 min compatible with inhibition of ROS generation due to O₂ depletion (Fig. 6). H₂DCF oxidation with simulated ischemia was markedly inhibited by preperfusion with DPI. H₂DCF oxidation also was markedly inhibited by the presence of catalase in the perfusate (i.e., extracellular). On the other hand, the addition of SOD did not decrease H₂DCF fluorescence, and the tracings suggest a more rapid rate of increase compatible with a more rapid dismutation of O· to H₂O₂ (Fig. 6). The decrease in DCF fluorescence with ischemia was only modestly inhibited (~18%) by 30-min preincubation of cells with 1 mM L-NAME, indicating that involvement of NO synthase (NOS) in DCF oxidation was relatively minor. There was no increase of DCF fluorescence in cells with simulated ischemia that had not been flow adapted (static cells).

View larger version (24K): [in this window] [in a new window]   Fig. 6.   Intracellular reactive O2 species (ROS) generation as detected by 2',7'-dichlorofluorscein (DCF) fluorescence. DCF fluorescence was measured at 530 nm (488 nm excitation) in a monolayer of flow-adapted BPAEC (5 dyn/cm2, 24 h) during simulated ischemia. Cells were prelabeled by perfusion with 5 µM 2',7'-dichlorodihydrofluorescein (H2DCF) diacetate. Conditions were the following: nonflow adapted (static) cells; cells perfused under control conditions; cells preperfused for 30 min with 20 µM DPI or 1 mM NG-nitro-L-arginine methyl ester (L-NAME); and addition of 25 units catalase (CAT) or of 20 units SOD to the perfusate.

Our previous studies documented that simulated ischemia causes [Ca²⁺]_i elevation in the pulmonary endothelium of the isolated lung (21). We characterized the kinetics of [Ca²⁺]_i changes in flow-adapted BPAEC during the ischemic episode using the fluorophore fura 2. The level of resting [Ca²⁺]_i in flow-adapted cells was ~135 nM. It began to increase ~30 s after flow cessation in flow-adapted, but not static, cells and reached a maximum level of 270 nM within 10 min (Fig. 7). The increase was markedly inhibited by the addition of EGTA (3 mM) to the perfusate. In contrast, preperfusion with the calmodulin inhibitor W-7 resulted in acceleration and elevation of amplitude of [Ca²⁺]_i response to ischemia in flow-adapted cells likely due to reduction in capacity of calmodulin to buffer [Ca²⁺]_i.

View larger version (23K): [in this window] [in a new window]   Fig. 7.   Intracellular Ca2+ measured by fura 2-acetoxymethyl ester (AM) fluorescence. Flow-adapted BPAEC were prelabeled with 5 µm fura 2-AM and fluorescence was measured during simulated ischemia (cessation of flow at time 0). Fluorescence at 510 nm was recorded in the ratiometric mode with excitation at 340 and 380 nm. A: real time kinetics of intracellular Ca2+ concentration ([Ca2+]i). These results were generated using a curve smoothing program (see MATERIALS AND METHODS). The curves on this panel correspond to: static (nonflow adapted) cells; control (flow-adapted) cells; 3 mM EGTA added to perfusate of flow-adapted cells; preperfusion of flow-adapted cells with 25 µM W-7. B: recording of the initial response (90 s after onset of ischemia) presented without curve smoothing and showing a resolution for this technique of ~5 s. The saw tooth pattern is produced by movement of the excitation monchromator in the ratio mode between 340 and 380 nm. The linear regression (straight line fit) of data before and after induction of simulated ischemia indicates a delay of ~30 s between the onset of ischemia and the rise of intracellular Ca2+. au, Arbitrary units.

Elevation of [Ca²⁺]_i is known to activate the endothelial NOS isoform via a calmodulin-dependent mechanism (16). We therefore studied NO generation during simulated ischemia in flow-adapted cells using detection of intracellular NO by a plasma membrane-permeable, NO-specific fluorescent dye, DAF-2 diacetate. Preincubation of BPAEC with 5 µM DAF-2 diacetate for 30 min resulted in trapping of deacetylated dye in the cells. This probe becomes highly fluorescent after reaction with NO (18). There was a linear increase in DAF-2 fluorescence in flow-adapted, but not static, cells that began ~50 s after flow cessation (Fig. 8) indicating intracellular NO generation. NO generation was abolished by the NOS inhibitor L-NAME and inhibited ~50% by the calmodulin inhibitor W-7. When the perfusate was saturated with 100% O₂, NO generation with ischemia remained linear during 5 min (Fig. 8).

View larger version (21K): [in this window] [in a new window]   Fig. 8.   Nitric oxide (NO) generation measured by diaminofluorescein (DAF-2) fluorescence. Flow-adapted BPAEC were prelabeled by perfusion with 5 µM DAF-2 diacetate. The curves indicate the following: static (nonflow adapted) cells with no additions; control cells (flow-adapted with no additions); cells perfused with O2-saturated perfusate; preincubation with 1 mM L-NAME; preperfusion for 30 min with 25 µM W-7.

Imaging of flow-adapted BPAEC preloaded with H₂DCF diacetate or DAF diacetate fluorescent probes in a fluorescence microscope showed that simulated ischemia results in a marked increase in cytoplasmic fluorescence intensity of both fluorophores (Fig. 9) as expected from the spectroscopic measurements.

View larger version (187K): [in this window] [in a new window]   Fig. 9.   Fluorescence microscopy for ROS and NO. Effect of simulated ischemia on fluorescent images of flow-adapted BPAEC loaded with H2DCF (A and B) or DAF (C and D). B and D were obtained with continuously perfused cells and A and C at 7 min after cessation of flow. The intense fluorescence in A and C indicates generation of ROS and NO, respectively, with simulated ischemia.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Endothelial responses to flow alterations via mechanosensor elements may represent a fundamental physiological paradigm. However, the response(s) to ischemia remain(s) to be more fully characterized and understood. In many settings, ischemia (flow cessation) leads to hypoxia and, therefore, the responses to mechanical signal(s) may become rapidly overshadowed by those related to a low O₂ level. Therefore, identification of the initial molecular and cellular events caused by flow cessation in endothelial cells represents an important and challenging goal. Our present study characterized the initial metabolic responses of flow-adapted endothelial cells to ischemia under conditions of adequate oxygenation. Adequate oxygenation was ensured by limiting observations to the initial several min of ischemia when PO₂ remained >45 mmHg, the physiological mixed venous PO₂. Adequacy of oxygenation during this period was confirmed by the demonstration of similar effects in air-saturated and O₂-saturated buffers.

Our previous studies using isolated, continuously ventilated rat or mouse lungs have shown that abrupt cessation of pulmonary perfusion causes rapid depolarization of the endothelial plasma membrane, generation of ROS, and increased [Ca²⁺]_i (1-3, 25). It is important to note that these effects are not a result of reperfusion and that anoxia/reoxygenation is not involved in this model because continuous ventilation maintains adequate oxygenation during lung ischemia and tissue ATP content does not change (3). Ischemia-mediated generation of ROS via a DPI-inhibitable oxidase(s) was confirmed with an in vitro system using flow-adapted BPAEC in artificial capillaries (27). Lung endothelium contains multiple DPI-inhibitable oxidases such as xanthine oxidase, mitochondrial NADH dehydrogenase, cytochrome P-450, and NO synthase that could contribute to ROS generation. Studies in perfused mouse lungs showed that "knock out" of gp91^phox eliminated the ROS response to ischemia providing evidence that a phagocyte-type NADPH oxidase is the responsible enzyme (2). Generation of O· outside the cell provides additional evidence that a plasma membrane-associated enzyme is responsible.

The present study utilized an original experimental model permitting direct spectroscopic study in real time or microscopic imaging of flow-adapted cells. Like the previous in vitro model using artificial capillaries (27), we found that adaptation to laminar flow was required to prime the response to simulated ischemia. The most likely mechanism for adaptation is upregulation of shear stress-sensitive elements during the adaptation period, although the specific sensor remains to be identified. Because cellular oxygenation is dependent on perfusate flow in the flow chamber that was utilized in the present study, this system approximates the effect of flow cessation in systemic vessels. Thus O₂ consumption by the cells led to a decrease in PO₂ to <45 ± 3 mmHg in about 4-5 min, although this period could be prolonged by perfusion with O₂-equilibrated buffer. Cellular changes with simulated ischemia were sufficiently rapid that they could be readily observed before O₂ limitation. Our results therefore show that endothelial cells adapted to flow as expected with vessels in vivo manifest an O₂-independent response to ischemia before the development of limiting hypoxia.

The complex response of endothelial cells to ischemia occurs only in flow-adapted cells and is characterized by a reproducible sequence of events. Within 15 s after flow cessation, depolarization of the plasma membrane occurs and is associated with increased generation of ROS as indicated by reduction of extracellular cyt c and by oxidation of the intracellular fluorescent indicator H₂DCF. Extracellular generation of O· is consistent with the presumed localization of NADPH oxidase to the plasma membrane as has been demonstrated for neutrophils and macrophages (13). Dismutation of extracellular O· perhaps catalyzed by extracellular SOD on the external plasma membrane (20) will generate H₂O₂. Subsequent diffusion of H₂O₂ into cells would lead to intracellular oxidation of H₂DCF, although other intracellular sources of ROS cannot be excluded. Increased [Ca²⁺]_i is a subsequent event (starting at about 30 s postischemia) followed by an increase in generation of NO (starting about 1 min postischemia). Thus the present study documents that simulated ischemia in flow-adapted BPAEC leads to endothelial plasma membrane depolarization and a respiratory burst followed by Ca²⁺ influx and increased generation of NO.

As noted above, O₂ consumption by the cells in the present experiments resulted in progressively decreasing O₂ availability. Phosphorescence of the medium was measured continuously during ischemia using the probe PdP1 that shows a detectable increase when PO₂ reaches a level of 45 ± 3 Torr. This enabled the calculation of O₂ consumption by the cells. Under standard conditions (37°C, P 760 mmHg, pH 7.4), the O₂ concentration in air-saturated perfusate (Krebs-Ringer bicarbonate) is 0.220 mM and a PO₂ of 45 mmHg corresponds to an O₂ concentration of ~0.066 mM. P is the normal pressure of 1 atmosphere. The volume of the chamber is 28 µl and contains ~10⁵ cells. Phosphorescence of PdP1 becomes detectable in ~5 min after induction of stimulated ischemia (Fig. 3). Assuming that ischemia results in a linear decrease of O₂ tension in the medium, the mean rate of O₂ consumption for the flow-adapted cells would be the following

The estimated O₂ consumption for flow-adapted cells in 100% O₂-saturated medium was approximately the same. The estimated mean value for O₂ consumption of static cells in the present experiments is 2.5 nmol/min per 10⁶ cells, which is in the same range as the value previously reported for endothelial cells from porcine thoracic aorta in static culture (8). The relative values for O₂ consumption by the cells is qualitatively apparent from the slopes of the phosphorescence plots versus time after medium PO₂ reached the point of phosphorescence increase (Fig. 4), although we considered quantitation more reliable using the earlier time points (between start of ischemia and initial increase in phosphorescence) when oxygenation was adequate.

Extracellular generation of O· was detected by reduction of the cell impermeable protein cyt c. Using Beers' law, we can calculate the concentration (C) of cyt c reduced by O· in the chamber during simulated ischemia. This value is equal to O· production because 1 mole of O· reduces 1 mole of cyt c
where the change in optical density (OD) between 0 and 2 min is 0.013 per min; the extinction coefficient of cyt c is 21.0 mM¹ · cm¹ (15); and the optical path length is 0.035 cm. Because the chamber volume is 28 µl and contains 10⁵ cells, the amount of O· produced during the initial 2 min of simulated ischemia by BPAEC is 5.1 nmol/min per 10⁶ cells. This may underestimate O· production because of rapid dismutation of O· (thus escaping the cyt c trap) and reoxidation of cyt c by H₂O₂. In the presence of catalase in the perfusate, the estimated rate of O· generation is ~6.1 nmol/min per 10⁶ cells. Thus O· generation accounts for ~70% of O₂ consumption by adapted BPAEC during ischemia, and the respiratory burst accounts for the entire increase in O₂ consumption for the flow-adapted, compared with static, cells. This estimated value for O· production by the confluent monolayer of flow-adapted BPAEC during simulated ischemia is similar to that seen during the respiratory burst of stimulated neutrophils (4-10 nmol/min per 10⁶ cells) (5, 9) and is markedly greater than seen with human umbilical vein endothelial cells simulated by interleukin-1 or recombinant interferon- (~0.4 nmol/min per 10⁶ cells) (24).

NADPH oxidase, as studied most extensively in phagocytes, is a multicomponent system consisting of both membrane-bound and cytosolic proteins. The rate-limiting step for NADPH oxidase activation is the assembly of these components on the inner leaf of the plasma membrane. In neutrophils, there is a significant lag period for O· generation after a stimulus such as digitonin (9). The rapid onset of O· generation after flow cessation (ischemia) in BPAEC suggests preassembly, at least in part, of the NADPH oxidase complex. Transmembrane electron transport via NADPH oxidase could couple with plasma membrane depolarization as observed with these cells, because transmembrane electron transfer provides an actual electrical current. Thus cessation of flow may trigger at least two fast membrane-associated events: inactivation of flow-sensitive K⁺ channels leading to membrane depolarization and an electron flux from cytosol to outside of the cells due to NADPH oxidase activation. These two fast-responding membrane-associated systems are coupled and might serve as the effector part of a mechanosensor and signaling mechanism for flow-adapted endothelial cells.

The present results show NO production during simulated ischemia with inhibition by L-NAME, suggesting activation of NOS as a secondary cellular response to flow cessation. The ~45-s delay between membrane depolarization and NO generation could reflect the time required for increase of [Ca²⁺]_i and subsequent activation of NOS through a Ca²⁺-dependent, calmodulin-mediated mechanism as indicated by results with EGTA and the calmodulin inhibitor W-7. The dependence of NO generation on Ca²⁺ and calmodulin suggest that endothelial NOS (eNOS) is the responsible isoform for the major fraction of NO production (23). A small amount of NO generation in the presence of EGTA could represent activation, in addition, of the Ca²⁺-independent NOS isoform previously reported by others (12). The channel responsible for Ca²⁺ influx in these cells remains to be identified. It is generally accepted that cultured endothelial cells do not have voltage-gated Ca²⁺ channels. However, voltage-gated Ca²⁺ channels have been described for freshly isolated capillary endothelial cells (6) suggesting their presence in vivo and raising the possibility that they are induced in endothelium during flow adaptation.

In summary, we have characterized the initial (0-2 min) response of endothelial cells to simulated ischemia during the period of adequate cellular oxygenation. Our results emphasize the fundamental difference in biology of endothelial cells flow adapted (the normal in vivo condition) or maintained in static culture (the artificial or pathological condition). Specifically, flow-adapted, but not static, endothelial cells are capable of sensing flow cessation and respond with ROS generation at an initial rate close to that of the respiratory burst in leukocytes. The subsequent responses include Ca²⁺ influx and increased generation of NO. Thus vascular endothelium can respond to the mechanical component of ischemia via complex biochemical signaling pathways.